A primary task of computer systems is to manage large quantities of information, generally referred to as data. The first computers typically stored data using off-line methods, for example by using punch cards and other primitive means. As built-in or on-line storage solutions became more affordable, data were instead stored in central memory banks. The first enterprise-wide computer systems consisted of central computers containing central data storage, and a large number of user terminals that accessed this server data by sending input and receiving output as characters to be displayed or printed at the terminal. Although these systems had a primitive user interface and data access became increasingly slower as the number of users grew, these systems nevertheless handled enterprise data with ease and great security.
The first servers, often referred to as mainframes or mini computers, ran on proprietary operating systems. Terminals usually had large input buffers where input was only checked against or committed to the server after entering text into a page or form. Many systems only displayed the character entered after it was received and confirmed by the server. Faster servers and more modern server operating systems, such as Unix and VMS, offered several advantages in that users could receive immediate feedback after each character was typed.
At the beginning of the 1980s decade, the growing popularity of microcomputers and personal workstations made it possible to store data locally. Enterprise data was distributed over networks of computer systems. To access information it was no longer necessary to have a continuous connection to central databases, and instead it was possible to copy information to a personal computer, edit and work with it, and then save it back to a file or database server later. Most microcomputers worked with data in logical chunks or files. This brought a lot of power to end users, but introduced problems in managing the large quantity of enterprise data that was no longer stored as a unique entity in one place. For example, a file that was being edited by one user could not usually be accessed or modified by other users at the same time. It was also difficult to manage multiple copies of the same data.
Toward the end of the 1980's faster microcomputers and networks made it practical to work with enterprise data in smaller chunks than files. One example of this new technology was the development of Structured Query Language (SQL) relational databases which made it possible to divide software programs into a ‘Client’ tier and a ‘Server’ tier, that communicated with each other over a network. Client-server computing thus made it possible to store information centrally, yet manage and work with it locally. In the client-server paradigm, the client systems concentrated on offering a user-friendly interface to server data, while the server systems were able to handle many client systems at once while safely managing enterprise data.
However, the increasing client-server computing introduced its share of problems. Protocols used to communicate between client and server became increasingly complex and difficult to manage. Enterprise IT departments needed increasingly greater resources to manage the proprietary implementations of client operating systems, server database systems and middleware protocols connecting the various ‘tiers’ of client-server systems. Data was no longer stored in one place but was required to be managed within a distributed network of systems. Client-server systems also lacked a major advantage of mainframes: in a client-server system any changes to the data on the server weren't immediately updated on the client.
Starting in the 1990s, the Internet has allowed businesses, organizations, and other enterprises to easily make information available to users without the complex architecture that client-server systems typically require. Today, an increasing number of software applications are moving their data and logic or functional processes back to the server tier, from which they can be accessed from the Internet by a wide variety of clients, including thin and very thin-clients, which typically consist of Internet browsers or small applications (applets) whose sole responsibility is providing an interface to the user. In many ways, Internet computing (often referred to as e-commerce) has brought back the data-handling advantages of mainframes. Within the e-commerce environment data that change on the server are immediately available to clients that access the data through the Internet (world-wide) or through an intranet (enterprise-wide).
Unfortunately, the rise of Internet commerce has also given rise to some of the disadvantages associated with mainframe technology. Most Internet connections that present data to the user or client process use the Hyper Text Transfer Protocol (HTTP) which is inherently “session-less.” This means that, for example, there is no totally reliable way for the server to automatically update the client display once the server data change. It also means that the server only checks the validity of the client or user input after the user sends back or submits an entire input form. This apparent disadvantage has also played an important role in the success of the Internet: because HTTP connections are session-less, they require much less processing power and much less memory on the server while the user is busy entering data. Thus, Internet applications running on web servers can be accessed by millions of people. Because HTTP and related Internet-based client-server systems do not provide continuous access to server data, systems sometimes incorporate lookup tables and pre-defined values that are cached locally. For example, a list of possible countries to be selected by a user of a web page can be sent to the user's computer when that page is first sent to the user and used thereafter for subsequent country selections. Client-server applications often pre-read the data from the server the moment an application or application window is opened, in order to present users with selection lists the moment they need them. This poses problems for data that frequently changes over time since the client system may allow users to select or enter data that is no longer valid. It also poses problems for large selection lists whose transmission to the client may take a long time.
To address this some systems incorporate a local cache of the data frequently accessed by the user. A web browser may, for example be configured to remember the last pages a user visited by storing them in a local cache file. A clear disadvantage of keeping such a local cache is that it is only useful as long as the user stays on the same client computer system. Also, the local cache may include references to web pages that no longer exist.
Some other systems with limited network bandwidth (like cell phones or personal organizers) can be deployed with built-in databases (such as dictionaries and thesauri), because it would be impractical to wait for the download of an entire database, which is needed before the data is of any use. This has the disadvantage that data stored in the device may no longer be up-to-date because it's really a static database. Also, the cost of cell phones and personal organizers is greatly increased by the need for megabytes of local storage. Another important consideration is that keeping valuable data in any local database makes it vulnerable to misuse and theft. What is needed is a mechanism that addresses these issues that allows a client-server system to retain some element of a session-based system, with its increase in performance, while at the same time offering a secure communication mechanism that requires little, if any, local storage of data.
Other attempts have been made to tackle some of the problems inherent with traditional computer system interfaces, and particularly with regard to user session administration and support. These attempts include the auto-complete function systems such as used in Microsoft Internet Explorer, the spell-as-you-go systems such as found in Microsoft Word, and the wide variety of client-server session managers such as Netopia's Timbuktu and Citrix Winframe.
Auto-Complete Functionality
Many current systems provide a mechanism to auto-complete words entered into fields and documents. This ‘auto-complete’ functionality is sometimes called ‘type-ahead’ or ‘predictive text entry’. Many web browsers such as Microsoft's Internet Explorer application will automatically ‘finish’ the entry of a URL, based on the history of web sites visited. E-mail programs including Microsoft Outlook will automatically complete names and e-mail addresses from the address book and a history of e-mails received and sent. Auto-completion in a different form is found in most graphical user interfaces, including operating systems such as Microsoft Windows and Apple Mac OS, that present lists to the user: When the user types the first character of a list entry, the user interface list will automatically scroll down to that entry. Many software development tools will automatically complete strings entered into program source code based on a known taxonomy of programming-language dependent key words and ‘function names’ or ‘class names’ previously entered by the developer. Some cell phones and personal organizers also automatically type-ahead address book entries or words from a built-in dictionary. Auto-complete functionality facilitates easy entry of data based on prediction of what options exist for the user at a single moment in time during entry of data.
Checking as You go
More and more word processing programs (most notably Microsoft Word and certain e-mail programs) include so-called ‘spell checking as you type’. These programs automatically check the spelling of words entered while the user is typing. In a way, this can be seen as ‘deferred auto-complete’, where the word processor highlights words after they were entered, if they don't exist in a known dictionary. These spell checking programs often allow the user to add their own words to the dictionary. This is similar to the ‘history lists’ that are maintained for the auto-completion of URLs in a web browser, except that in this case the words are manually added to the list of possible ‘completions’ by the user.
Software Component Technologies
Software component technologies have provided a measure of component generation useful in client/server systems. One of these technologies is OpenDoc, a collaboration between Apple Computer, Inc. and IBM Corporation (amongst others) to allow development of software components that would closely interact, and together form applications. One of the promises of OpenDoc was that it would allow small developers to build components that users could purchase and link together to create applications that do exactly what the users want, and would make existing ‘bloat-ware’ applications (notably Microsoft Office and Corel's WordPerfect Office/Corel Office) redundant, but the technology was dropped several years ago in favor of newer technologies such as CORBA (Common Object Request Broker Architecture), developed by the Object Management Group to allow transparent communication and interoperability between software components.
Object-oriented languages and even non-object-oriented (database) systems have used component technologies to implement technical functionality. The NeXTstep operating system from NeXT Computer, Inc. (which was later acquired by Apple Computer, Inc. and evolved into the Mac operating system Mac OS X) had an object-oriented architecture from its original beginnings, that allowed software developers to create applications based on predefined, well-tested and reliable components. Components could be ‘passive’ user interface elements (such as entry fields, scroll areas, tab panes etc) used in application windows. But components could also be active and show dynamic data (such as a component displaying a clock, world map with highlight of daylight and night, ticker tape showing stock symbols, graphs showing computer system activity, etc.). The NeXT operating system used object frameworks in the Objective C language to achieve its high level of abstraction which is needed for components to work well. Later, Sun Microsystems, Inc. developed the Java language specification in part to achieve the same goal of interoperability. To date, Java has probably been the most successful ‘open’ (operating system independent) language used to build software components. It is even used on certain web sites that allow ‘Java applets’ on the user's Internet browser to continuously show up-to-date information on the client system.
WebObjects, an object-oriented technology developed by Apple Computer, Inc. is an Internet application server with related development tools, which was first developed by NeXT Computer, Inc. WebObjects uses object oriented frameworks that allow distribution of application logic between server and client. Clients can be HTML-based, but can also be Java applets. WebObjects uses proprietary technology that automatically synchronizes application objects between client and server. The layer that synchronizes data objects between the client and the server is called the ‘Enterprise Object Distribution’ (EODistribution), part of Apple's Enterprise Objects Framework (EOF), and is transparent to the client software components and the server software components.
Session Management
Both Netopia's Timbuktu remote access systems, and Citrix, Inc.'s Winframe terminal server product, allow some element of remote access to server applications from a client system. These products synchronize user data and server data, transparently distributing all user input to the server and return all server (display) output to the client. Timbuktu does this with very little specific knowledge about the application and operating system used. This allows it to transparently work on both Microsoft Windows and Mac OS platforms. Technologies similar to Timbuktu do exist and perform the same kind of ‘screen sharing’. For example, the Virtual Network Computing (VNC) system is one example of an open source software program that achieves the same goals and also works with Linux and Unix platforms.
Citrix Winframe has taken the same idea a step further by incorporating intimate knowledge of the Microsoft Windows operating system (and its Win32 APIs) to further optimize synchronization of user input and application output on the server. It can then use this detailed knowledge of the Microsoft Windows APIs to only redraw areas of the screen that it knows will change based on a user action: for example, Winframe may redraw a menu that is pulled down by the user without needing to access the server application because it knows how a menu will work.
Software Applications
Several application providers have also built upon these technologies to provide applications and application services of use to the end-user. These applications include computer-based thesauri, on-line media systems and electronic encyclopediae.
The International Standards Organization (as detailed further in ISO 2788-1986 Documentation—Guidelines for the Establishment and Development of monolingual thesauri and ISO 5964-1985 Documentation—Guidelines for the Establishment and Development of multilingual thesauri) determines suggested specifications for electronic thesauri, and thesaurus management software is now available from numerous software vendors world-wide. However, most systems have clear limitations that compromise their user-friendliness. Most commonly this is because they use a large third-party database system, such as those from Oracle Software, Inc. or Informix, Inc. as a back-end database. This means that any thesaurus terms that are displayed to the user are fetched from the database and then presented in a user interface. If one user changes the contents of the thesaurus, other users will only notice that change after re-fetching the data. While of little concern in small or infrequently changing environments, this problem is a considerable one within larger organizations and with rapidly updated content changes, for example in media publishing applications when thesaurus terms are being linked to new newspaper or magazine articles. This type of work is usually done by multiple documentalists (media content authors) simultaneously. To avoid ‘mixing up’ terms linked to articles, each documentalist must be assigned a certain range of articles to ‘enrich’ (which in one instance may be the act of adding metadata and thesaurus terms to a document). Clearly, in these situations there is a great need for live updates of data entered by these users, but a similar need exists for all client-server database programs.